Solving Semidefinite Programs Research Articles

Near-term quantum algorithm for solving the MaxCut problem with fewer quantum resources

MaxCut is an NP-hard combinatorial optimization problem in graph theory. The quantum approximate optimization algorithms (QAOAs) offer new methods for solving such problems, which are potentially better than classical schemes. However, the requirement of N qubits for solving graphs with N-vertex in QAOA, coupled with the large-N trainability issue due to barren plateaus, poses a substantial challenge for noisy intermediate-scale quantum (NISQ) devices. Recently, a hybrid quantum–classical algorithm has been proposed for solving semidefinite programs (SDPs), named NISQ SDP Solver (NSS). In this paper, we study the performance of the NSS for solving MaxCut problem via the introduction of a near-term quantum algorithm and execution of experimental simulations. Since the MaxCut problem admits a relaxation into an SDP formulation, the SDP can be solved using NSS, with a subsequent classical post-processing step converting the hybrid density matrix into a MaxCut solution. After performing these steps, the near-term quantum algorithm will also inherit the advantages of NSS. We analyze the algorithm as compared to QAOAs and find that the depth of the quantum circuit is independent of the number of edges on the graph. Our algorithm requires logN qubits and has O(1) circuit depth. This implies that it uses exponentially fewer qubits compared to QAOAs while also requiring a significantly reduced circuit depth to solve the MaxCut problem. To demonstrate the effectiveness of the NSS, we focused on 3-regular graphs, 9-regular graphs, and ER graphs. Numerical results indicate that the quantum algorithm achieves a comparable approximation ratio with classical methods by solving the original high dimensional problem within a lower dimensional ansatz space. Analysis under various initial states shows that the algorithm exhibits a remarkably stable performance.

On symmetry adapted bases in trigonometric optimization

The problem of computing the global minimum of a trigonometric polynomial is computationally hard. We address this problem for the case, where the polynomial is invariant under the exponential action of a finite group. The strategy is to follow an established relaxation strategy in order to obtain a converging hierarchy of lower bounds. Those bounds are obtained by numerically solving semi-definite programs (SDPs) on the cone of positive semi-definite Hermitian Toeplitz matrices, which is outlined in the book of Dumitrescu Dumitrescu (2007). To exploit the invariance, we show that the group has an induced action on the Toeplitz matrices and prove that the feasible region of the SDP can be restricted to the invariant matrices, whilst retaining the same solution. Then we construct a symmetry adapted basis tailored to this group action, which allows us to block-diagonalize invariant matrices and thus reduce the computational complexity to solve the SDP.The approach is in its generality novel for trigonometric optimization and complements the one that was proposed as a poster at the ISSAC 2022 conference Hubert et al. (2022) and later extended to Hubert et al. (2024). In the previous work, we first used the invariance of the trigonometric polynomial to obtain a classical polynomial optimization problem on the orbit space and subsequently relaxed the problem to an SDP. Now, we first make the relaxation and then exploit invariance.Partial results of this article have been presented as a poster at the ISSAC 2023 conference Metzlaff (2023).

Sum-of-squares certificates for Vizing's conjecture via determining Gröbner bases

The famous open Vizing conjecture claims that the domination number of the Cartesian product graph of two graphs G and H is at least the product of the domination numbers of G and H. Recently Gaar, Krenn, Margulies and Wiegele used the graph class G of all graphs with nG vertices and domination number kG and reformulated Vizing's conjecture as the problem that for all graph classes G and H the Vizing polynomial is sum-of-squares (SOS) modulo the Vizing ideal. By solving semidefinite programs (SDPs) and clever guessing they derived SOS-certificates for some values of kG, nG, kH, and nH.In this paper, we consider their approach for kG=kH=1. For this case we are able to derive the unique reduced Gröbner basis of the Vizing ideal. Based on this, we deduce the minimum degree (nG+nH−1)/2 of an SOS-certificate for Vizing's conjecture, which is the first result of this kind. Furthermore, we present a method to find certificates for graph classes G and H with nG+nH−1=d for general d, which is again based on solving SDPs, but does not depend on guessing and depends on much smaller SDPs. We implement our new method in SageMath and give new SOS-certificates for all graph classes G and H with kG=kH=1 and nG+nH≤15.

Fast implementation for semidefinite programs with positive matrix completion

Solving semidefinite programs (SDPs) in a short time is the key to managing various mathematical optimization problems. The matrix-completion primal–dual interior-point method (MC-PDIPM) extracts a sparse structure of input SDP by factorizing the variable matrices. In this paper, we propose a new factorization based on the inverse of the variable matrix to enhance the performance of MC-PDIPM. We also use multithreaded parallel computing to deal with the major bottlenecks in MC-PDIPM. Numerical results show that the new factorization and multithreaded computing reduce the computation time for SDPs that have structural sparsity.

A Novel Approach for Solving Semidefinite Programs

A novel linearizing alternating direction augmented Lagrangian approach is proposed for effectively solving semidefinite programs (SDP). For every iteration, by fixing the other variables, the proposed approach alternatively optimizes the dual variables and the dual slack variables; then the primal variables, that is, Lagrange multipliers, are updated. In addition, the proposed approach renews all the variables in closed forms without solving any system of linear equations. Global convergence of the proposed approach is proved under mild conditions, and two numerical problems are given to demonstrate the effectiveness of the presented approach.

A trust region method for solving semidefinite programs

When using interior point methods for solving semidefinite programs (SDP), one needs to solve a system of linear equations at each iteration. For problems of large size, solving the system of linear equations can be very expensive. In this paper, we propose a trust region algorithm for solving SDP problems. At each iteration we perform a number of conjugate gradient iterations, but do not need to solve a system of linear equations. Under mild assumptions, the convergence of this algorithm is established. Numerical examples are given to illustrate the convergence results obtained.

A robust algorithm for semidefinite programming

Current successful methods for solving semidefinite programs (SDPs) are based on primal–dual interior-point approaches. These usually involve a symmetrization step to allow for application of Newton's method followed by block elimination to reduce the size of the Newton equation. Both these steps create ill-conditioning in the Newton equation and singularity of the Jacobian of the optimality conditions at the optimum. In order to avoid the ill-conditioning, we derive and test a backwards stable primal–dual interior-point method for SDP. Relative to current public domain software, we realize both a distinct improvement in the accuracy of the optimum and a reduction in the number of iterations. This is true for random problems as well as for problems of special structure. Our algorithm is based on a Gauss–Newton approach applied to a single bilinear form of the optimality conditions. The well-conditioned Jacobian allows for a preconditioned (matrix-free) iterative method for finding the search direction at each iteration.

An Interior Point Method for Solving Semidefinite Programs Using Cutting Planes and Weighted Analytic Centers

We investigate solving semidefinite programs (SDPs) with an interior point method called SDP‐CUT, which utilizes weighted analytic centers and cutting plane constraints. SDP‐CUT iteratively refines the feasible region to achieve the optimal solution. The algorithm uses Newton’s method to compute the weighted analytic center. We investigate different stepsize determining techniques. We found that using Newton′s method with exact line search is generally the best implementation of the algorithm. We have also compared our algorithm to the SDPT3 method and found that SDP‐CUT initially gets into the neighborhood of the optimal solution in less iterations on all our test problems. SDP‐CUT also took less iterations to reach optimality on many of the problems. However, SDPT3 required less iterations on most of the test problems and less time on all the problems. Some theoretical properties of the convergence of SDP‐CUT are also discussed.

A Randomized Method for Solving Semidefinite Programs

Proposed is a novel iterative method for solving semidefinite programs. It exploits the ideology of cutting hyperplane through the center of mass of a convex body. To estimate the center of mass, we use a random walk technique known as the Hit-and-Run algorithm. The results of numerical simulations are compared to those obtained with presently available approaches. Robust versions of the method are considered, where the coefficient matrices contain norm-bounded uncertainties.

A nonlinear programming algorithm for solving semidefinite programs via low-rank factorization

In this paper, we present a nonlinear programming algorithm for solving semidefinite programs (SDPs) in standard form. The algorithm's distinguishing feature is a change of variables that replaces the symmetric, positive semidefinite variable X of the SDP with a rectangular variable R according to the factorization X=RR T . The rank of the factorization, i.e., the number of columns of R, is chosen minimally so as to enhance computational speed while maintaining equivalence with the SDP. Fundamental results concerning the convergence of the algorithm are derived, and encouraging computational results on some large-scale test problems are also presented.

Approximation of the Stability Number of a Graph via Copositive Programming

Lovasz and Schrijver [SIAM J. Optim., 1 (1991), pp. 166--190] showed how to formulate increasingly tight approximations of the stable set polytope of a graph by solving semidefinite programs (SDPs) of increasing size (lift-and-project method). In this paper we present a similar idea. We show how the stability number can be computed as the solution of a conic linear program (LP) over the cone of copositive matrices. Subsequently, we show how to approximate the copositive cone ever more closely via a hierarchy of linear or semidefinite programs of increasing size (liftings). The latter idea is based on recent work by Parrilo [Structured Semidefinite Programs and Semi-algebraic Geometry Methods in Robustness and Optimization, Ph. D. thesis, California Institute of Technology, Pasadena, CA, 2000]. In this way we can compute the stability number $\alpha(G)$ of any graph $G(V,E)$ after at most $\alpha(G)^2$ successive liftings for the LP-based approximations. One can compare this to the $n - \alpha(G)-1$ bound for the LP-based lift-and-project scheme of Lovasz and Schrijver. Our approach therefore requires fewer liftings for families of graphs where $\alpha(G) < O(\sqrt{n})$. We show that the first SDP-based approximation for $\alpha(G)$ in our series of increasingly tight approximations coincides with the $\vartheta'$-function of Schrijver [IEEE Trans. Inform. Theory, 25 (1979), pp. 425--429]. We further show that the second approximation is tight for complements of triangle-free graphs and for odd cycles.

Infinite-dimensional convex optimization in optimal and robust control theory

Many engineering problems can be shown to be equivalent to solving semidefinite programs (SPs), i.e., convex optimization problems involving linear matrix inequalities (LMIs). Powerful computation tools are available for such problems in the finite-dimensional case. However, the problems arising in optimal and robust control theory are often infinite dimensional, and so adequate computation tools are not available. The key to tackling such problems with finite computation tools is to have a primal-dual formulation of the problem without duality gap. In this paper we study infinite-dimensional SPs and present a lifting technique to recast SPs as parameterized linear programs (LPs). This enables the wealth of theoretical tools available for infinite-dimensional LPs to be extended to infinite-dimensional SPs. In particular, we develop some new sufficient conditions for the lack of a duality gap for infinite-dimensional SPs and give an exact characterization of the primal and dual problems for these cases. Both primal and dual problems are formed as infinite-dimensional SP problems, with finite truncations to each giving upper and lower bounds, respectively, on the exact solution to the infinite-dimensional problem. Thus, these results can form the basis of practical computation schemes for infinite-dimensional problems, which require only finite-dimensional computation tools. To illustrate the power of these tools we apply the results to two previously unsolved optimization problems, namely minimizing the l/sub 1/ norm of a closed-loop system subject to bounds on the frequency response magnitude at a finite number of points and/or bounds on the H/sub 2/ norm.